Allow parsing cast expressions.

Author: reitermarkus

src/expr.rs

@@ -29,8 +29,8 @@ use std::ops::{
 use crate::literal::{self, CChar};
 use crate::token::{Kind as TokenKind, Token};
 use crate::ToCexprResult;
-use nom::branch::alt;
-use nom::combinator::{complete, map, map_opt};
+use nom::branch::{alt, permutation};
+use nom::combinator::{complete, map, map_opt, opt};
 use nom::multi::{fold_many0, many0, separated_list0};
 use nom::sequence::{delimited, pair, preceded};
 use nom::*;
@@ -54,6 +54,7 @@ pub enum EvalResult {
     Float(f64),
     Char(CChar),
     Str(Vec<u8>),
+    Cast(Vec<Vec<u8>>, Box<Self>),
     Invalid,
 }
 
@@ -126,6 +127,10 @@ fn identifier_token(input: &[Token]) -> CResult<'_, &[u8]> {
     }
 }
 
+fn keyword(c: &'static str) -> impl Fn(&[Token]) -> CResult<'_, &[u8]> {
+    exact_token!(Keyword, c.as_bytes())
+}
+
 fn p(c: &'static str) -> impl Fn(&[Token]) -> CResult<'_, &[u8]> {
     exact_token!(Punctuation, c.as_bytes())
 }
@@ -289,6 +294,10 @@ where
     nom::combinator::map_opt(f, EvalResult::as_numeric)
 }
 
+fn expr_cast(input: (Vec<Vec<u8>>, EvalResult)) -> EvalResult {
+  EvalResult::Cast(input.0, Box::new(input.1))
+}
+
 impl<'a> PRef<'a> {
     fn unary(self, input: &'_ [Token]) -> CResult<'_, EvalResult> {
         alt((
@@ -473,6 +482,10 @@ impl<'a> PRef<'a> {
 
     fn expr(self, input: &'_ [Token]) -> CResult<'_, EvalResult> {
         alt((
+            map(
+                pair(|i| self.cast(i), |i| self.expr(i)),
+                expr_cast
+            ),
             |i| self.numeric_expr(i),
             delimited(p("("), |i| self.expr(i), p(")")),
             |i| self.concat_str(i),
@@ -482,6 +495,88 @@ impl<'a> PRef<'a> {
         .to_cexpr_result()
     }
 
+    fn cast(self, input: &'_ [Token]) -> CResult<'_, Vec<Vec<u8>>> {
+        delimited(p("("), |i| self.ty(i), p(")"))(input)
+    }
+
+    fn int_ty(input: &'_ [Token]) -> CResult<'_, Vec<&[u8]>> {
+      fn int_signedness(input: &'_ [Token]) -> CResult<'_, &[u8]> {
+          alt((
+              keyword("unsigned"),
+              keyword("signed"),
+          ))(input)
+      }
+
+      fn int_longness(input: &'_ [Token]) -> CResult<'_, &[u8]> {
+          alt((
+              keyword("short"),
+              keyword("long"),
+          ))(input)
+      }
+
+      alt((
+          // [const] [(un)signed] long long [int]
+          map(
+            permutation((opt(keyword("const")), opt(int_signedness), keyword("long"), keyword("long"), opt(keyword("int")))),
+            |(_, s, i1, i2, _)| if let Some(s) = s {
+              if s == b"signed" {
+                vec![i1, i2]
+              } else {
+                vec![s, i1, i2]
+              }
+            } else {
+              vec![i1, i2]
+            },
+          ),
+          // [const] [(un)signed] long/short [int]
+          map(
+              permutation((opt(keyword("const")), opt(int_signedness), int_longness, opt(keyword("int")))),
+              |(_, s, i, _)| if let Some(s) = s {
+                if s == b"signed" {
+                  vec![i]
+                } else {
+                  vec![s, i]
+                }
+              } else {
+                vec![i]
+              },
+          ),
+          // [const] [(un)signed] char/int
+          map(
+              permutation((opt(keyword("const")), opt(int_signedness), alt((keyword("char"), keyword("int"))))),
+              |(_, s, i)| if let Some(s) = s {
+                if s == b"signed" && i == b"int" {
+                  vec![i]
+                } else {
+                  vec![s, i]
+                }
+              } else {
+                vec![i]
+              },
+          ),
+      ))(input)
+    }
+
+    fn ty(self, input: &'_ [Token]) -> CResult<'_, Vec<Vec<u8>>> {
+      map(
+        alt((
+          // [const] <identifier>
+          map(
+              permutation((opt(keyword("const")), identifier_token)),
+              |(_, id)| vec![id],
+          ),
+          // [const] bool
+          map(
+              permutation((opt(keyword("const")), keyword("bool"))),
+              |(_, b)| vec![b],
+          ),
+          Self::int_ty,
+        )),
+        |v| v.into_iter().map(|t| t.to_vec()).collect(),
+      )(input)
+      .to_cexpr_result()
+    }
+
     fn macro_definition(self, input: &'_ [Token]) -> CResult<'_, (&'_ [u8], EvalResult)> {
         pair(identifier_token, |i| self.expr(i))(input)
     }

tests/clang.rs

@@ -12,6 +12,7 @@ use std::collections::HashMap;
 use std::io::Write;
 use std::str::{self, FromStr};
 use std::{char, ffi, mem, ptr, slice};
+use std::num::Wrapping;
 
 use cexpr::assert_full_parse;
 use cexpr::expr::{fn_macro_declaration, EvalResult, IdentifierParser};
@@ -25,18 +26,17 @@ fn test_definition(
     tokens: &[Token],
     idents: &mut HashMap<Vec<u8>, EvalResult>,
 ) -> bool {
+    use cexpr::expr::EvalResult::*;
+
     fn bytes_to_int(value: &[u8]) -> Option<EvalResult> {
-        str::from_utf8(value)
-            .ok()
-            .map(|s| s.replace("n", "-"))
-            .map(|s| s.replace("_", ""))
-            .and_then(|v| i64::from_str(&v).ok())
-            .map(::std::num::Wrapping)
+        let s = str::from_utf8(value).ok()?;
+        let s = s.rsplit_once('_').map(|(_, s)| s).unwrap_or(s);
+
+        i64::from_str(&s.replace("n", "-")).ok()
+            .map(Wrapping)
             .map(Int)
     }
 
-    use cexpr::expr::EvalResult::*;
-
     let display_name = String::from_utf8_lossy(&ident).into_owned();
 
     let functional;
@@ -127,6 +127,7 @@ fn test_definition(
                 return false;
             }
         }
+
         assert_full_parse(IdentifierParser::new(&fnidents).expr(&expr_tokens))
     } else {
         IdentifierParser::new(idents)

tests/input/fail.h

@@ -1,4 +1,3 @@
-#define FAIL_function_like(x) 3
 #define FAIL_empty
 #define FAIL_invalid_for_radix 0b2
 #define FAIL_shift_by_float 3<<1f

tests/input/int_signed.h

@@ -1,3 +1,5 @@
 #define Int_n3 -(-(-3))
 #define Int_n5 -3-2
 #define Int_n9223372036854775808 -9223372036854775808
+
+#define Fn_Int_n9(_3) _3*-3

tests/input/int_unsigned.h

@@ -10,7 +10,7 @@
 #define Int_124 124u
 #define Int_125 125uL
 #define Int_126 126LuL
-#define Int_16 (((1)<<4ULL))/*comment*/ 
+#define Int_16 (((1)<<4ULL))/*comment*/
 #define Int_13 1|8^6&2<<1
 
 #define Int_47 32|15
@@ -27,3 +27,4 @@
 #define Int_n9223372036854775808 9223372036854775808
 
 #define Fn_Int_9(_3) _3*3
+#define Fn_Int_unused_arg_3(x_unused_0) 3